Chemical kinetics modelling of combustion processes in SI engines

The need for improving the efficiency and reducing emissions is a
constant challenge in combustion engine design. For spark ignition
engines, these challenges have been targeted in the past decade or
so, through ‘engine downsizing’ which refers to a reduction in engine
displacement accompanied by turbocharging. Besides the benefits of
this, it is expected to aggravate the already serious issue of engine
knock owing to increased cylinder pressure. Engine knock which is
a consequence of an abnormal mode of combustion in SI engines, is
a performance limiting phenomenon and potentially damaging to the
engine parts. It is therefore of great interest to develop capability to
predict autoignition which leads to engine knock. Traditionally, rather
rudimentary skeletal chemical kinetics models have been used for autoignition
modelling, however, they either produce incorrect predictions
or are only limited to certain fuels. In this work, realistic chemical
kinetics of gasoline surrogate oxidation has been employed to address
these issues.
A holistic modelling approach has been employed to predict combustion,
cyclic variability, end gas autoignition and knock propensity
of a turbocharged SI engine. This was achieved by first developing
a Fortran code for chemical kinetics calculations which was
then coupled with a quasi-dimensional thermodynamic combustion
modelling code called LUSIE and the commercial package, GT-Power.
The resulting code allowed fast and appreciably accurate predictions
of the effects of operating condition on autoignition. Modelling was
validated through comparisons with engine experimental data at all
stages.
Constant volume chemical kinetics modelling of the autoignition of
various gasoline surrogate components, i.e. iso-octane, n-heptane,
toluene and ethanol, by using three reduced mechanisms revealed
how the conversion rate of relatively less reactive blend components,
toluene and ethanol, is accelerated as they scavenge active radical
formed during the oxidation of n-heptane and iso-octane. Autoignition
modelling in engines offered an insight into the fuel-engine interactions
and that how the composition of a gasoline surrogate should
be selected. The simulations also demonstrated the reduced relevance
of research and motor octane numbers to the determination of gasoline
surrogates and that it is crucial for a gasoline surrogate to reflect
the composition of the target gasoline and that optimising its physicochemical
properties and octane numbers to match those of the gasoline
does not guarantee that the surrogate will mimic the autoignition
behaviour of gasoline.
During combustion modelling, possible deficiencies in in-cylinder turbulence
predictions and possible inaccuracies in turbulent entrainment
velocity model required an optimisation of the turbulent length
scale in the eddy burn-up model to achieve the correct combustion
rate. After the prediction of a correct mean cycle at a certain engine
speed, effects of variation in intake air temperature and spark timing
were studied without the need for any model adjustment. Autoignition
predictions at various conditions of a downsized, turbocharged
engine agreed remarkably well with experimental values. When coupled
with a simple cyclic variability model, the autoignition predictions
for the full spectrum of cylinder pressures allowed determination
of a percentage of the severely autoigniting cycles at any given spark timing or intake temperature. Based on that, a knock-limited spark advance was predicted within an accuracy of 2° of crank angle.